Like most investors, we at True are obviously excited by the direct applications of AI.
We have made several investments in the space already including Jitty (AI to find your dream home), SeeChange (AI to prevent theft in retail stores) and Eyva (AI-driven insights for product portfolios), to name a few. And we’ll keep looking for more.
But alongside direct applications, we are just as interested in the second- and third-order consequences of increased adoption of AI amongst consumers.
We hear a lot about how AI is threatening incumbents. Reference sites like TripAdvisor and Yelp are grappling with traffic erosion from ‘zero click searches’, where content is summarised into AI answers without a monetisable click-through to underlying websites. Dominant players in travel are racing to combat disintermediation risk posed by agents that promise direct bookings. Even Google Search - arguably one of the greatest monopolies of all time - is under threat from Open AI, Perplexity and other challengers.
A theme we’re thinking through is how transaction flows on the Internet will change in an era where AI agents - rather than human users - are taking on responsibility for large portions of the online purchasing journey. We’ve written before that agentic commerce needs to solve some fundamental challenges before it truly takes off - and even then, won’t be suited to all circumstances. But it seems inevitable that everyday consumers will soon be accompanied by agentic AI handling at least some portion of their online shopping.
Attention was all you need
Agentic commerce changes things dramatically because the organising principle of Internet commerce for the past two decades has been the scarcity of human attention.
As tech penetrates more and more of our daily lives, online services are locked in an intense battle for attention and engagement. Accordingly, Internet commerce has been shaped for humans’ finite attention spans. Websites and apps are designed to capture interest quickly, nudge purchasing decisions through slick UX, visual merchandising and behavioural cues, and optimise conversion whilst driving maximum value from each interaction. Every step is carefully curated for a world with humans low on attention and patience.
But the paradigm is shifting with agentic commerce.
When the robots take over
To state the obvious: software is very different to humans.
It has dramatically higher capacity for sustained attention and dramatically lower search costs. While most human users in most situations are looking for the path of least resistance, often using heuristics and gut feel to make decisions, software can ruthlessly, surgically and tirelessly pursue the ‘optimal’ route given a set of defined parameters.
No longer is mastery (and manipulation) of human behaviour the key to success for Internet commerce - in fact, it might soon become irrelevant in a world where attention is no longer scarce.
We see this impacting a few different areas of today’s transactional experience on the Internet:
Interfaces: Rich interfaces with thoughtful visual merchandising help humans navigate complex websites quickly, using the inherent information density of images (vs. words) to steer attention and trigger emotional responses. That will need to evolve. Even if the end-state sees agents navigating graphical user interfaces like humans (vs. plugging in directly an API or MCP server), they will be more single-minded in their efforts to solve a defined problem and therefore less susceptible to eye-catching merchandising seeking to distract. Best-in-class design will look very different.
Personalisation: An everpresent buzzword, today’s ‘personalisation’ software often relies on inferring needs and wants rather than expecting a user to explicitly lay them out. Proxies for intent come from the source of traffic (e.g. a particular ad creative), behaviour patterns in website navigation, first/third party data, and so on. By contrast, AI agents equipped with the relevant user context for a defined task could make personalisation much more simple with less guesswork - a straightforward exchange between the agents and the website, like a two-way personal shopping experience. “I’ll have your team speak to my team,” and all that. Check out Fabric, who are building part of the data stack to make that a reality.
Product Information: Most product display pages feature thoughtfully curated emotive imagery and well-crafted copy designed for emotional human users. But agents will be more sceptical and rigorous. Expect to see deeper information and more objectivity. For example, consider performance nutrition: agents will disregard subjective marketing terms like “natural”, “wholesome” and “healthy” alongside images of models with six-packs. Instead, we’ll trend towards new objective product claims. Forward-thinking brands are already pushing in that direction, like David’s Calories from Protein (“CFP”) score.
Pricing: Price discrimination across different channels is an age-old tactic to optimise margin whilst capturing as much demand as possible. Historically, this has been possible because the friction and marginal search cost to find the best deal is too much for most human users. But an AI agent acting on my behalf can scan every corner of the Internet to find the cheapest way of purchasing a given item. Pricing structures could go either way depending on category: simplicity and consistency across all channels at one end, or individualised through AI driven negotiation bots like Nibble at the other.
Upsell: If your order-level unit economics rely heavily on impulse upsell in the checkout flow, it may be time to rethink. I think and hope endless upsell screens will die, and I certainly won’t miss clicking through offers for at-seat food, car rental, public transport tickets, travel insurance, and so on when trying to buy a flight ticket. These unsolicited offers won’t survive the agentic UX paradigm - no consumer will want to be badgered with follow-up asks from their AI agent. Tactics designed for emotional and impulsive human users simply won’t make sense anymore.
Payments: No merchant wants to fall at the final hurdle. The need to offer consumers a fast and low-friction method to pay has led to an ever-expanding universe of payment options: credit card, debit card, Apple/Google Pay, buy-now-pay-later, one-click, ‘Pay by Bank’, and so on. Speed is key to avoid losing the human’s attention and losing the conversion. But if an agent can and will be more patient, they may start optimising for different things beyond speed alone. Expect smarter digital wallets with loyalty/offer optimisation, value sharing with merchants, group purchasing, etc. - an era of consumer-facing payments orchestration might be starting (check out Google’s AP2 announcement for some initial thoughts on how payments are evolving for agentic commerce).
There is no doubt that many more areas will change fundamentally once human attention (or lack thereof) is no longer the organising principle of Internet commerce. That’s what makes it such an exciting time to be building and investing for the new reality.
We are excited to back founders reimagining how it all works, so if you have a vision for how interfaces, purchasing journeys and transactions evolve in the era of AI, please get in touch here.